METHODS

Altogether, questionnaires were mailed to 21 201 men and women aged 16–64 years chosen at random from the age-sex registers of 163 GPs in 34 general practices across Britain and to 993 randomly selected members of the armed forces. The practices were chosen to give a broad geographical coverage and to ensure that industries with known exposure to vibration (a primary focus of inquiry) were represented in the sample, while members of the armed forces (who are cared for by military rather than civilian doctors) were sampled additionally. A single reminder was sent to non-responders about five weeks later. Details of the sampling methods, questionnaire, and validity of the measuring instruments have been published previously.^1,17,18

Among other things, enquiries were made about self reported hearing difficulties, tinnitus, and the wearing of a hearing aid; the total time spent working in noisy places where there was a need to shout to be heard (which roughly equates to noise exposures exceeding 85–90 dB(A)¹⁹); and smoking habits.

Difficulty in hearing was assessed by the question: “How well can you hear a person who is talking to you when he is sitting on your right [left] side in a quiet room?”. This was modelled on an item from the MRC National Survey of Hearing, in which the response category “moderate difficulty” corresponded to a measured mean hearing impairment of about 45 dB.²⁰ We classed subjects as severely affected if they wore a hearing aid or reported severe hearing difficulty in the worse affected ear; as being moderately affected if they reported moderate hearing difficulty in the worse ear; and as normal if they reported only slight hearing difficulty in the worse ear or no difficulty at all. (In those who only reported on hearing in one ear, the category was that for the ear on which information was given.) Tinnitus was identified by asking: “During the past 12 months have you had noises in your head or ears (such as ringing, buzzing or whistling) which lasted longer than five minutes?” It was defined as “persistent” if it was reported to occur most or all of the time. A smoker was defined as someone who had smoked at least once a day for a month or longer, and classed according to whether he or she still smoked regularly (current smoker) or did not (former smoker).

Information was also collected on age and sex as potential confounding factors, and on complaints of frequently feeling tired or stressed or suffering frequent headaches, as possible markers of lowered threshold for reporting symptoms in general. All statistical analyses were carried out in STATA, release 5.²¹ Associations of hearing impairment and tinnitus with smoking were examined by logistic regression, with adjustment for potential confounding. Separate analyses were conducted according to duration of employment in a noisy occupation. The findings were expressed as prevalence ratios (PRs) with associated 95% confidence intervals (CIs). These were derived from the corresponding odds ratios using a formula proposed by Zocchetti and colleagues.²²

RESULTS

Questionnaires were returned by 12 907 subjects (58% of those selected for study), with a higher response rate in older subjects and women. The occupational distribution of responders was generally similar to that in the preceding national census.²³ Further details on the response pattern have been published elsewhere.¹⁷

The respondents included 3184 current smokers, 3329 former smokers, and 6394 lifelong non-smokers. Non-smokers tended to be younger and reported working fewer years in a noisy occupation than ex- and current smokers.

Among 10 418 respondents who provided details on hearing, 9759 were classed as having slight or no hearing difficulty, 348 as having moderate hearing difficulty, and 311 as having severe hearing difficulty (including 165 who wore a hearing aid). Persistent tinnitus was reported by 527 subjects. As reported previously,¹ the prevalence of tinnitus was higher in those with hearing difficulties (about twice as common in age standardised comparisons), and both symptoms were more common at older ages. Subjects who felt frequently tired or stressed, or suffered frequent headaches, also reported such problems more often.

Table 1 shows the association of hearing difficulties and tinnitus with smoking habit and years in noisy work, after adjustment for report of tiredness, stress, and headaches, and for age and sex. The baseline for comparison was subjects who were free of auditory problems, and all risks were estimated relative to those of non-smokers who had never worked in a noisy job.

The risks of moderate-to-severe hearing difficulty, severe hearing difficulty, and persistent tinnitus increased with years spent in noisy work for all categories of smoking habit, being some three to five fold higher in those employed for more than five years than in those never employed in noisy work.

Within strata of noise exposure, ex- and current smokers had a moderately higher risk of hearing difficulty than lifetime non-smokers in most comparisons. For example, the PR for moderate-to-severe hearing difficulty was 1.5 (95% CI 1.1 to 2.1) in current versus never-smokers without exposure to noise at work. By contrast, after allowance for age and duration of noisy employment, smoking status was only weakly related to tinnitus if at all.

The combined effect on hearing difficulty of current smoking and long term noise exposure was generally less than multiplicative, and more consistent with an additive effect. Thus, among subjects with moderate-to-severe hearing difficulty, the risk difference between smokers and non-smokers was similar in every stratum of years employed in noise; and the same was true among subjects with severe hearing difficulty who had never worked in a noisy job or had been employed in it for up to five years. In those with severe hearing difficulty who had been employed for more than five years in a noisy job, the risk difference between smokers and non-smokers was greater, but the interaction was still less than multiplicative.

DISCUSSION

This population based survey tends to confirm that smoking is associated with hearing difficulty, and that this relation exists even in those who have never been exposed to occupational sources of noise. However, it indicates that any such effect is modest in comparison with long term employment in a noisy job, and that the combined effects of smoking and noise are additive rather than multiplicative.

A strength of our survey was the large, geographically dispersed, and well defined sampling base, which included over 22 000 randomly selected subjects across England, Wales, and Scotland. The distribution of occupations in respondents was nationally representative, being similar to that at the preceding national census.

Participation was incomplete (response rate 58%), but response bias would only explain the association of smoking with hearing difficulty if smokers with symptoms participated more readily than smokers without. This seems unlikely, and similar associations were found in those who responded to the questionnaire at the first invitation and those who required a reminder (data not presented), providing evidence against this pattern of bias.

In view of the relatively small number of working aged subjects reporting hearing loss, smoking status rather than pack-years of smoking was used as the index of exposure. This precluded examination of a dose-response relation, but subjects are less likely to have been misclassified using this metric of exposure. Major error arising from the question on hearing impairment also seems unlikely, as the validity of this item has previously been established.²⁰ However, our metric of exposure to noise (years worked in a noisy job where there was a need to shout to be heard) was crude to the extent that it did not fully reflect intensity of noise above a likely threshold of 85–90 dB(A). Among noise exposed workers, differences in intensity of noise exposure or degree of compliance with wearing of ear defenders may have existed by smoking status. This may have caused residual confounding after allowance for years of employment in noise, but cannot explain the increased risk in smokers who had never held a noisy job.

A number of other potential explanations for the association need to be considered. Smokers may have reported hearing difficulty more readily than non-smokers, perhaps as part of a general tendency to report somatic symptoms at a lower threshold; but the association persisted despite adjustment for markers of such a tendency (frequent complaints of headaches, tiredness, and stress).

Alternatively, assuming a genuine effect, smoking may act as a direct ototoxin;²⁴ may impair blood flow to the cochlea by inducing vasospasm, atherogenic narrowing, and thrombotic occlusion of the nutrient arteries;^5,25 or may alter blood viscosity,^15,26 or generate carboxyhaemoglobin,^27,28 exacerbating any problems of local hypoxia.

There have been only a few population based surveys of smoking and hearing loss previously. When 1662 elderly subjects from the Framingham study underwent audiometric screening, no association was found between hearing thresholds and current smoking status or number of cigarettes smoked.¹¹ And hearing thresholds in middle age did not deteriorate more rapidly during follow up among smokers than non-smokers from rural Denmark.¹⁴ But in the US Health Interview Survey,¹⁰ men who smoked two or more packs per day more often reported hearing problems than non-smokers; in a population based cross-sectional survey based in Beaver Dam, USA,¹³ a 25 dB hearing loss in the worse ear was 1.7 times more common in current smokers than lifetime non-smokers after adjustment for other factors; and in another study, which sampled non-smokers from the community, a moderately increased risk of hearing loss (OR 1.3) was found in association with exposure to environmental tobacco smoke.²⁹

Surveys set within programmes of health screening and hearing conservation have produced some of the most convincing evidence that smoking is detrimental to hearing. For example, in Japan, where labour law mandates periodic audiometry for all employees irrespective of noise exposure, two large surveys found worse hearing among smokers than non-smokers in non-noisy white collar occupations, and a dose-response effect among current smokers which was much clearer for high frequency hearing loss than for low frequency loss.^3,9 Mizoue and colleagues⁹ have suggested that inner ear cells responsible for high frequency hearing are more vulnerable to ischaemic damage, as they are located at the end of nutrient arteries, and that the inconsistency of findings across studies may reflect failure to distinguish the different potential patterns of hearing loss.

The combined effects of noise and smoking have only occasionally been explored. In keeping with our findings, Mizoue et al reported an additive effect, based on periodic health data from 4624 Japanese steel company workers.⁹ But by contrast, Prince et al found that the combined effect of these risk factors was more than that from simple additivity.¹⁶

Workers should be encouraged as a matter of good public health policy to avoid the hazards of both noise and smoking. But our results suggest that from the viewpoint of hearing there is no strong imperative to limit the employment of smokers to a greater extent than non-smokers where ambient noise levels are significant. Counselling may be appropriate in these circumstances, but our data indicate only a small additional risk from smoking and underscore the importance of noise as the primary target for preventive measures in hearing conservation programmes.